Distributed backstepping based control of multiple UAV formation flight subject to time delays
Distributed backstepping based control of multiple UAV formation flight subject to time delays
复制标题
DOI：
10.1049/iet-cta.2019.1151
复制
发表时间：
2020-08-13
影响因子：
2.6
通讯作者：
Lewis, Frank
中科院分区：
文献类型：
--
作者：
Kartal, Yusuf;Subbarao, Kamesh;Lewis, Frank
在这项研究中，作者提出了一种基于反步法的分布式编队控制方法，该方法在多架无人机（UAV）通信存在时延的情况下依然稳定。无人机的集中式编队控制要求每个智能体与其他智能体保持一定的间隔距离，这给无人机的通信网络带来了负担。为克服这一问题，作者考虑了一种分布式控制方案，其中每个智能体根据从其邻居收集到的状态信息来更新自身的姿态和位置。他们没有直接控制螺旋桨产生的推力，而是将无人机的数学模型划分为两个子系统，一个是线性姿态控制回路，另一个是非线性位置控制回路。然后设计了一个基于反步法的外部位置控制器，它与级联控制系统的内部姿态控制器无缝衔接。在分布式网络时延的影响下，通过严格的李雅普诺夫 - 克拉索夫斯基分析确定了闭环稳定性。利用有向图拓扑结构和分布式反步结构，表明稳定性判据与时延无关。所提出的控制算法在仿真中得到验证，然后在硬件中实现，实际飞行测试实验证明了这些算法的有效性。
In this study, the authors propose a backstepping-based, distributed formation control method that is stable independent of time delays in communication among multiple unmanned aerial vehicles (UAVs). Centralised formation control of UAVs requires each agent to maintain a separation distance from other agents, which burdens the communication network of the UAVs. To overcome this problem, the authors consider a distributed control scheme wherein each agent updates its attitude and position based on the state information gathered through its neighbours. Instead of directly controlling the thrust generated by the propellers, they partition the mathematical model of the UAV into two subsystems, a linear attitude control loop and a non-linear position control loop. A backstepping-based outer position controller is then designed that interfaces seamlessly with the inner attitude controller of the cascaded control system. The closed-loop stability is established using a rigorous Lyapunov-Krasovskii analysis under the influence of distributed network time delays. Using the directed graph topology and a distributed backstepping structure, it is shown that the stability criterion is delay-independent. The proposed control algorithms are verified in simulation and then implemented in hardware, and actual flight test experiments prove the validity of these algorithms.